Potential Biases Introduced by Conflating Screening and Diagnostic Testing in Colorectal Cancer Screening Surveillance.

Abstract

Screening and postsymptomatic diagnostic testing are often conflated in cancer screening surveillance research. We examined the error in estimated colorectal cancer screening prevalence due to the conflation of screening and diagnostic testing. Using data from the 2008 National Health Interview Survey, we compared weighted prevalence estimates of the use of all testing (screening and diagnostic) and screening in at-risk adults and calculated the overestimation of screening prevalence across sociodemographic groups. The population screening prevalence was overestimated by 23.3%, and the level of overestimation varied widely across sociodemographic groups (median, 22.6%; mean, 24.8%). The highest levels of overestimation were in non-Hispanic white females (27.4%), adults ages 50-54 years (32.0%), and those with the highest socioeconomic vulnerability [low educational attainment (31.3%), low poverty ratio (32.5%), no usual source of health care (54.4%), and not insured (51.6%); all P < 0.001]. When the impetus for testing was not included, colorectal cancer screening prevalence was overestimated, and patterns of overestimation often aligned with social and economic vulnerability. These results are of concern to researchers who use survey data from the Behavioral Risk Factor Surveillance System (BRFSS) to assess cancer screening behaviors, as it is currently not designed to distinguish diagnostic testing from screening. Surveillance research in cancer screening that does not consider the impetus for testing risks measurement error of screening prevalence, impeding progress toward improving population health. Ultimately, to craft relevant screening benchmarks and interventions, we must look beyond "what" and "when" and include "why."